具有线性趋势的残差自回归模型的估计方法

发布时间：2018-03-03 09:17

本文选题：线性趋势　切入点：残差自回归　出处：《统计与决策》2016年11期 　论文类型：期刊论文

【摘要】：为了探寻具有线性趋势的残差自回归模型的较为合适的估计方法,文章以残差AR(2)模型为例,对直接最小二乘法、两步法、非线性最小二乘法和化归法进行了Monte Carlo模拟,拟合和预测结果显示非线性最小二乘法和化归法的均方误差和平均绝对误差相同且最小。此外,还利用1980—2013年河南省人均GDP经济数据进行了拟合与预测实证分析,得到了与模拟比较相类似的结果,这说明非线性最小二乘法和化归法是较优的估计方法。进一步地,基于非线性最小二乘法,给出了河南省人均GDP的短期预测。
[Abstract]:In order to find a more suitable method for estimating the residual autoregressive model with linear trend, Monte Carlo simulation of direct least square method, two-step method, nonlinear least square method and regression method is carried out in this paper, taking the residuals ARQ2) model as an example. The fitting and forecasting results show that the mean square error and mean absolute error of the nonlinear least square method and the regression method are the same and the minimum. In addition, the empirical analysis of fitting and forecasting is carried out by using the economic data of Henan Province per capita GDP from 1980 to 2013. The results are similar to those obtained by the simulation, which shows that the nonlinear least square method and the reduction method are the better estimation methods. Furthermore, based on the nonlinear least square method, the short-term prediction of GDP per capita in Henan Province is given.
【作者单位】：河南科技大学数学与统计学院;
【基金】：国家社会科学基金资助项目(11BTJ017) 河南省国际科技合作计划项目(134300510034) 河南科技大学博士科研启动基金项目(09001322)
【分类号】：F830;F224

【相似文献】